PDF to Excel Converter
Extract tables and text from a PDF into a .xlsx spreadsheet, one sheet per page. Runs entirely in your browser — files are never uploaded to a server.
Table detection here is a best-effort position-based estimate, not true table structure recognition — it works well on simple, clearly-aligned tables and may misalign complex tables with merged cells or irregular spacing.
Click to upload or drag and drop
Select a PDF file
How it works
Each page's text is grouped into rows by vertical position, then into columns by clustering nearby horizontal positions — reconstructing a grid without any real knowledge of the PDF's original table structure. This works well for clean, simple tables and plain text, but isn't a substitute for true table recognition on complex documents. For scanned PDFs with no text layer, the tool falls back to OCR on its own. If you only need plain text rather than a spreadsheet, Scanned PDF to Text (OCR) is the dedicated tool.
How rows and columns are decided
Text fragments are sorted from top to bottom. A new row starts when a fragment's vertical position differs from the previous fragment's by more than 3 points. The comparison is fragment to fragment, so items at heights 700, 698.5 and 696.9 all join one row, while one at 690 starts the next. For columns, the left edge of every fragment on the page is sorted, and a new column starts whenever the gap to the previous edge is over 15 points. Each fragment then goes into the nearest column.
A worked example
Three lines of text with left edges at 72, 250 and 330 points give three columns: Item, Qty, Price, then Pens, 12, 3.50, then Ink, 4, 12.25. Right-aligned amounts that start at 318, 330 and 336 points stay in one column because each gap is under 15 points, so "1,200.00" and "45.00" land together. Two fragments starting 13 points apart, such as at 72 and 85, are treated as one column and end up in one cell as "X Y".
What ends up in the workbook
Each page becomes its own sheet named Page 1, Page 2 and so on, and a page with no text becomes a sheet with one empty cell. The file is named like invoice-extracted.xlsx. Every cell is written as text, so values such as 12 and 3.50 arrive as text and may need converting to numbers in Excel. Columns are worked out for the whole page, so headings or paragraphs above a table take part in the clustering and can split its columns.
Scanned PDFs
If no page contains selectable text, the tool switches to OCR on its own, in English, and a note on the result screen says so. Each recognized line becomes one row with a single column, because OCR gives no column positions.
Frequently asked questions
How accurate is the table detection?
It's a position-based estimate, not true table structure recognition — reliable for simple, clearly-aligned tables, but it can misalign complex tables with merged cells or irregular spacing.
Does this use OCR for scanned PDFs?
Yes, automatically. It reads the PDF's text layer first, and if no page has any selectable text it switches to English OCR instead and shows a note on the result screen.
Why are my numbers treated as text in Excel?
Every extracted value is written to the sheet as text. Convert the column to numbers in Excel if you want to add them up or sort them numerically.
Why did my table come out with too many columns?
Columns are found by clustering the left edges of all text on the page. Left-aligned text, headings and notes spread out over many positions can create extra columns, so cleaner, evenly aligned tables convert best.
What do I get from a scanned PDF?
The tool falls back to OCR automatically when the PDF has no selectable text. Each line becomes one row in a single column, since OCR cannot recover where table columns were.